Centralized Bazaar Setup With No Network Connection

I am new to Bazaar and am trying to set up a system to support website development by a 2 man team. We operate in an MS Windows environment where we each have a development pc which has internet access, but no local network access. We also have a desktop pc, for which we don't have admin access and don't plan to install Bazaar, which also has internet access as well as local network access. These pc's run XP. On the network we have a test server which cannot access the internet, but is accessible over the network, there is currently no ftp/sftp access to it, we use remote desktop and a shared folder on its hdd to access it. Finally, our production server is accessible both via the web and over the network, we connect to this using ftp and remote desktop. Both servers run Win 2003.

Having played around with Bazaar a little I think the set up we need is:

1. On the development desktop pcs we set up a directory to hold our local master copy of the website. We then checkout/branch from this, copying the site to the directory pointed to by the pc's webserver. Development of the site is then done in the webserver copy so we can test it.

Question: What's the difference between a branch and a checkout, I haven't quite grasped that yet?

Once we complete a change we use Send to generate a patch which contains the changes made in the webserver branch versus our local master copy. The master branch/webserver branch is required because Send doesn't appear to be able to generate a patch based on revisions within a single branch, it works across a pair of branches.

Question: Am I correct about Send, or have I just not figured out the correct parameters to work on a single branch?

Having generated a patch this can be physically copied onto a usb stick and moved to our second local pc which has network access.

2. The networked test server will have the same set up as the development pc, a local master branch, and a checkout/branch from this in the directory pointed to by the webserver. The dual set up here allows us to develop emergency fixes if the need arises.

When we generate a patch on the local development pc it will be physically copied onto the networked local pc, copied onto the test servers shared directory and from there we can pull/merge it into the webserver branch of the website. After successful testing and user signoff we can then pull/merge the change into the server master branch.

Question: I'm not clear on the difference between pull/merge, they both appear to be able to process a patch, when should I use which?

If we make an emergency fix the 2 branch set up will enable us to generate a patch which can be used to updated the production server code.

3. Finally, having signed off a change the patch file originally created on the local development pc can be copied over to the production server. The production server will only have one branch which will be located in the directory pointed to by the web server. We will then pull/merge the patch into the production website code to install the update.

Comments anyone? The lack of network access makes the process clumsy and we're going to need to be careful with merging patches to ensure that revisions are consistent across all our machines. Any suggestions for a more automated set up would be welcome. Note that we are not part of IT, we're regarded as business users so we have restricted rights and access, there are many hoops to jump through to get server ports opened etc. We also deal with client personal data so there are plenty of data security requirements which restrict what access is available too. The combination of the two has driven the enviroment design that's outlined above.

Thanks in advance!